CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   OpenFOAM Running, Solving & CFD (https://www.cfd-online.com/Forums/openfoam-solving/)
-   -   how to use DES to capture vortex shedding in turbulence (https://www.cfd-online.com/Forums/openfoam-solving/106772-how-use-des-capture-vortex-shedding-turbulence.html)

 lancjin September 7, 2012 14:01

DES for unsteady flow (vortex shedding) in turbulence

1 Attachment(s)
Hi everyone,

I'm a new comer here (i'm really pretty new since i just spent half an hour trying to explain my problem to be posted and when i clicked it's said i was logged out.......:(), and also new comer in OF. I played with Fluent for some cases and now i just switched to OF, which has more freedom to control all kinds of things.

well, go back to my question, i'm trying to use DES to simulate the vortex shedding over a inclined flat plate(not cylinder, but still vortex shedding exists), inside of a 2D rectangular computational domain. Re based on the panel length is 20 000, two main references are:

2001_J_Turbulence_Separated flow around a flat plate at high incidence an LES investigation

2003_I_J_Num_Methods in Fluids_Comparison of DES RANS LES for separated flow around a flat plate at high incidence

I know the LES and DES are developed for 3D case, but could i just use 2D, which is easier and faster, to get some qualititive structures, such as periodical phenomenon first? After checking my mesh and boundary conditions on 2D, then i could switch to 3D....

The boundary conditions are: (for up and down, symmetryPlane is used for all variables, and front and back, just empty for 2D case)
0/U file
internalField: uniform 0
inlet: fixedValue uniform (0.3 0 0)
outlet: convectiveOutlet(which code is found in this forum, developed by Takuya OSHIMA 05/Sep/2007), convectiveVelocity uniform 0.3, ddtScheme backward;
panel1&panel2(two sides of the flat plate): fixedValue uniform 0

0/p file
internalField: uniform 0
outlet: fixedValue uniform 0

0/nuSgs file
internalField uniform 0
outlet: convectiveOutlet, convectiveVelocity uniform 0.3, ddtScheme backward

0/nuTilda file
internalField uniform 1.0e-10
inlet: fixedValue uniform 0
outlet: convectiveOutlet, convectiveVelocity uniform 0.3; ddtScheme backward
panel1&2: fixedValue uniform 0

PisoFoam is used and the time discretization is backward second order. And the time step is just increased from Delta_t = 0.001s to 0.005s(CFLmax is around 3.). Since this simulation is based on implicit scheme and pressure residual is well controlled, so the 'large' time step should not be a problem here. And after the computational time t = 40 s, there is still no periodical phenomenon shown yet...What could be wrong here? could anyone give any advice or instructions? i've been struggling with this for weeks since the simulation test on single computer, even with 6 cores, is too slow to get any result....I will really appreciate your help!!

 lancjin September 9, 2012 13:02

http://imageshack.us/photo/my-images/210/paneli.jpg
The geometry is shown as above .